基于地统计学的土壤污染源解析模型差异对比

Differences of soil pollution source analysis models based on geostatistics

为探究不同源解析模型的适用性,以松阴溪流域土壤重金属为例,分别采用主成分-绝对主成分-多元线性回归模型(PCA-APCS-MLR)、正定矩阵因子分解模型(PMF)以及UNMIX模型对研究区土壤重金属进行污染源解析,并运用地统计学方法对3种模型计算的贡献度进行插值分析,最后分析对比源解析结果。结果表明:对于研究区土壤污染源来说,通过PCA-APCSMLR模型可识别出自然源、工业源和交通源,而通过PMF模型、UNMIX模型分析均识别出了交通-农药-污灌源、自然源以及工业源。将3种模型结合分析可知,自然源是导致研究区土壤Cd、Pb、As、Cu、Zn污染的主要因素,总贡献率为30.1%;Cr、Ni污染的主要因素是工业源,总贡献率为37.4%;Hg污染主要与交通源、农药-污灌源有关,总贡献率分别为18.0%、14.5%。研究表明,PCAAPCS-MLR模型在判别主要污染源类型时具有一定优势,而PMF模型、UNMIX模型在计算源贡献率时较为准确,在源解析时需要将不同模型结合起来,使其结果更加准确。

To explore the applicability of different source analysis models,using soil heavy metals in Songyin Creek as an example,we analyzed the pollution sources in the study area using PCA-APCS-MLR,PMF,and UNMIX models.We used interpolation analysis to determine the contribution of the three models using geostatistics.Finally,we compared the source analysis results.The results showed that for soil pollution sources in the study area,natural sources could be identified by the PCA-APCS-MLR model,and traffic-pesticide pollution irrigation sources,natural and industrial sources could be identified by the PMF and UNMIX models.According to these three models,natural sources were the main sources of soil Cd,Pb,As,Cu,and Zn,and the total contribution rate was 30.1%;Cr and Ni pollution were from industrial sources,and the total contribution rate was 37.4%;Hg pollution was mainly related to traffic sources and pesticide pollution irrigation sources with rate of 18.0%and 14.5%,respectively.This study indicates that the PCA-APCS-MLR model has certain advantages in distinguishing between the types of major pollution sources,while the PMF and UNMIX models are more accurate when calculating the source contribution rate.Different models should be combined in the source analysis to increase the accuracy of the results.